Testing for a unit root in ERM exchange rates in the presence of structural breaks: evidence from the bootstrap

Abstract

This paper explores the extent to which accounting for structural breaks in ERM exchange rates affects inferences on the presence of a unit root in these exchange rates. Four ERM exchange rates, found by previous empirical studies to be nonstationary, are examined. In contrast to previous empirical studies, multiple structural breaks are allowed for to account for multiple realignments in the central parities of these exchange rates. Bootstrapped critical values, personalized to the pattern of breaks of each exchange rate, are used for statistical inference. Consistent with the theoretical conclusion by Froot and Obstfeld (1991), the results suggest that all four ERM exchange rates are stationary. Therefore, accounting for breaks in ERM exchange rates does affect inferences on the presence of a unit root in these exchange rates.